Disclosed is a computer program stored in a computer-readable storage medium for achieving the above-described objects. When the computer program is executed by one or more processors, the computer program causes the one or more processors to perform the following operations of generating video data for diagnosing a health condition and a pathological symptom on the basis of a biosignal. The operations include receiving a biosignal of a user, preprocessing the biosignal to generate a plurality of pieces of video sub-data, and selecting at least two of the plurality of pieces of video sub-data to generate one or more pieces of video data.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
8. The method of claim 7, wherein the deep learning model is generated when the one or more processors input each of the training input datasets to the one or more network functions, calculate errors by comparing each piece of output data computed through the one or more network functions with each of the training output datasets that each correspond to labels of the training input datasets, adjust weights of the one or more network functions in a backpropagation manner on the basis of the errors, determine whether to stop training the one or more network functions using verification data when the training is performed for a predetermined epoch or more, and test performance of the one or more network functions using a test dataset to determine whether to activate the one or more network functions.
The invention relates to a method for training and validating a deep learning model using network functions. The method addresses the challenge of efficiently training and testing deep learning models to ensure optimal performance before deployment. The process involves inputting training datasets into one or more network functions, which process the data to generate output. The output is compared to corresponding labeled training datasets to calculate errors. These errors are used to adjust the weights of the network functions through backpropagation, iteratively refining the model. Training continues for a predetermined number of epochs, after which verification data is used to assess whether training should stop. Once training is complete, the model's performance is tested using a separate test dataset to determine if it meets activation criteria. This method ensures that the deep learning model is accurately trained and validated before being deployed for real-world applications. The approach optimizes model accuracy and reliability by systematically adjusting weights and validating performance against independent datasets.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
January 11, 2021
May 28, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.